Review of Literature
The focus of national policy research has traditionally been the individual, the service provider, or entire service systems. Though the breadth and depth of research on families and family care is enormous (Ramey, Krauss & Simeonsson, 1989; Turnball, Beegle & Stowe, 2002), it has been framed primarily in terms of family process and dynamics such as functioning and transition planning (Smith, 1997) -- rarely in terms of policy planning or advocacy. The research plan is predicated on two suppositions: (1) the family household is the most important system of (informal) support in the U.S., and (2) that the strength and vitality of the family household can be greatly facilitated by elevating the visibility of the family in disability policy and services planning.
Aging represents a critical research issue in family demographics. The cohort of children born during the two decades after World War II – “baby boomers” is unprecedented in size and has stimulated research on aging with a disability as well as initiatives in the areas of healthy aging and secondary conditions. To what extent do the households and their members represent an alternative "system" of support? To what extent will a generational shift of the magnitude suggested by the baby-boom generation impose unprecedented demands on the long-term care system? To what extent can family households be supported as an alternative to an already overextended long-term care and support system? A demographic analysis of the family household of persons with a disability will begin this important dialogue.
Variability in the support of persons with disabilities across states is the norm rather than exception and data at the national level provides only the grossest portrait for policy makers at the state level. This represents a dilemma for understanding the status of persons with disabilities using national data sets since implementation and funding is largely under the control of state government. State-level disability data is notoriously difficult to access. The proposed project is a pilot analysis using the decennial Census Long Form (LF) data, which collected both disability and family household information (albeit limited in scope). Using this data set, the project will explore establishing state-level summaries of persons with disabilities and the structure of their family households.
Sample Population and Methodology
The “sample” will be drawn from the 2000 LF Public Use data files based on the long-form questionnaire given to 1 in 6 Americans. The project will make use of the 5% Public Use Files (the other option is a 1% sample, representative at the national level). We propose to analyze ten states, one from each of the federal regions of the U.S.
Data from the 5% file has not yet been released but is anticipated during the 4th quarter of 2003. The LF Files contain basic demographics such as households and family status, race, ethnicity, ages of household members, marital status, grandparents as caregivers, citizenship status, and educational attainment, veteran status, disability, employment status, occupation, income, and poverty status. Housing items include housing totals, urban and rural, household size, value of home, and monthly shelter costs. Disability questions on the long form census were limited to three types of queries: (1) the presence of visual or hearing impairments; (2) whether a condition limits basic physical activities, and (3) whether a person had chronic, long-term problems in learning, dressing, going outside or working.
Data Analysis
Family coding schemes in the LF focus on the form of familial relationship or if none exists, the form of relationship among the unrelated members of the household. In addition, the coding identifies the household “head” in terms of ownership or under whose name the rental is registered. Though a wide range of family types could be extracted from these coding schemes, three basic households will be employed: (1) persons living alone, (2) unrelated housemates, and (3) family households. Subjects will be collapsed into five basic ethnic/racial groups: White, Black, Hispanic, Asian, or Native American. The Hispanic category will include individuals of any race with Central or South American ethnic identification.
Baby boomers are defined as persons born between 1946 and 1964. Three family groups will be the focus of the analysis: (1) Family Household, Non-Head: Persons with disabilities from the post war generation living in family households; (2) Family Household, Head: Persons with disability from the post war generation living as either the head of household or spouse household head; and (3) Non-Family Household: Persons with disability from the post war generation living in non-family households (either alone or with unrelated roommates).
The first group, Family Household, Non-Head is the primary focus of interest in this analysis. While the notion of living as a “dependent” within a household can only be approximated with the available data, the point of the analysis will be to estimate the size and character of homes within which other family members play critical support roles.Estimates will be based on population weights associated with each household and individual within household. Weights are provided within the data sets by the Bureau of the Census. Since variance estimation in standard statistical software underestimates sample variances for data based on complex sample designs, variance estimates were constructed using Sudaan 7.5 software (Shah, Barnwell & Bieler, 1997), which includes adjustments for cluster-correlated data. These adjusted variance estimates will be employed in developing confidence intervals for testing population differences in prevalence and frequency. Standard errors will be used to develop 95% confidence intervals to assess the magnitude of differences in estimates across the data sets.
Basic socio-economic demographics of age, gender, educational achievement, occupation, household income, and physical structure of home will be summarized for the household head and member with a disability and compared across the different types of families. There are no explicit codes indicating the family member with a disability is in some manner dependent upon support from other family members. By definition, families exist in co-dependent relationships regardless of the presence of an impairment or significant limitation in activity. We will use two criteria for imputing “support” by other household members who do not have a disability: (1) citing “need for assistance” in an activity in conjunction with (2) absence of economic independence. Status on this latter variable will be based on the earned income or other income sources of the person with a disability (exclusive of public assistance) falling below the Federally designated poverty level for someone living alone in 2000.
Total estimated populations, proportions, and averaged values will be compared across the main subgroups defined in the research questions (e.g., types of family households, racial/ethnic groups). In all comparisons we will attempt to anticipate adjustments for demographic variables that may exaggerate or obscure differences due to underlying relationships with disablement. For example, comparisons across poor vs. not poor would need age adjustments since age is so closely associated with disablement. The challenge to the analysis will be working with a sample size that allows only a limited number of adjustments: we cannot control for everything. The plan will be to evaluate differences across the basic strata and make adjustments accordingly.